> This blog post will show how a fix for XFree86 and linuxthreads ended up causing a major threading regression about 7 years after the fix was created.
> The regression was in `pthread_create`. Thread creation performed very slowly as the number of threads in a process increased. This bug is present on CentOS 5.3 (and earlier) and other linux distros as well.
Actually, while it merits a (fine) blog post, it is far from being the craziest as the original post title suggests. But what I find really important is surely the following paragraph:
> It is also very possible that this bug impacted research done before August 15, 2008 (in the best case because Linux distro releases are slow) on building high performance threaded applications.
Research is hard since you may be affected with every possible cause unknown to you.
I thought it was pretty crazy that a tweak to prevent XFree86 from breaking on 64bit systems + support for the ELF small code execution model would later result in pthread_create overhead so large that creating threads on a P4 would be as slow as creating threads on a 386.
> This blog post will show how a fix for XFree86 and linuxthreads ended up causing a major threading regression about 7 years after the fix was created.
> The regression was in `pthread_create`. Thread creation performed very slowly as the number of threads in a process increased. This bug is present on CentOS 5.3 (and earlier) and other linux distros as well.
Actually, while it merits a (fine) blog post, it is far from being the craziest as the original post title suggests. But what I find really important is surely the following paragraph:
> It is also very possible that this bug impacted research done before August 15, 2008 (in the best case because Linux distro releases are slow) on building high performance threaded applications.
Research is hard since you may be affected with every possible cause unknown to you.